Structure Fires in Laboratories Marty Ahrens Fire Analysis and Research Division National Fire Protection Association February 2016 National Fire Protection Association, 1 Batterymarch Park, Quincy, MA 02169-7471 www.nfpa.org
Acknowledgements The National Fire Protection Association thanks all the fire departments and state fire authorities who participate in the National Fire Incident Reporting System (NFIRS) and the annual NFPA fire experience survey. These firefighters are the original sources of the detailed data that make this analysis possible. Their contributions allow us to estimate the size of the fire problem. We are also grateful to the U.S. Fire Administration for its work in developing, coordinating, and maintaining NFIRS. For more information about the National Fire Protection Association, visit www.nfpa.org or call 617-770- 3000. To learn more about the One-Stop Data Shop go to www.nfpa.org/osds or call 617-984-7451. Copies of this analysis are available from: National Fire Protection Association One-Stop Data Shop 1 Batterymarch Park Quincy, MA 02169-7471 www.nfpa.org e-mail: osds@nfpa.org phone: 617-984-7451 NFPA Index No. Copyright 2016, National Fire Protection Association, Quincy, MA This custom analysis is prepared by and copyright is held by the National Fire Protection Association. Notwithstanding the custom nature of this analysis, the NFPA retains all rights to utilize all or any part of this analysis, including any information, text, charts, tables or diagrams developed or produced as part hereof in any manner whatsoever as it deems appropriate, including but not limited to the further commercial dissemination hereof by any means or media to any party. Purchaser is hereby licensed to reproduce this material for his or her own use and benefit, and to display this in his/her printed material, publications, articles or website. Except as specifically set out in the initial request, purchaser may not assign, transfer or grant any rights to use this material to any third parties without permission of NFPA.
Structure Fires in Laboratories The National Fire Incident Reporting System (NFIRS), the source of detailed information about fires reported to local fire departments, identifies laboratories in two ways. Property use 629 is used to identify laboratory or science laboratory occupancies. Fires in these properties may or may not start in laboratory areas. Table 1 shows that an average of 142 fires per year were reported in laboratory properties in 2009-2013, resulting in an average of nine civilian injuries and $14 million in direct property damage annually. No deaths were reported in these properties during this period. Almost half (45%) of the fires in these properties started in laboratory areas. Table 2 shows that during the same period, an average of 247 reported structure fires per year started in laboratory areas in properties of all types. These fires caused an average of 24 civilian injuries and $13 million in direct property damage annually. The average number of deaths rounded to zero per year. More than one-third (35%) of fires starting in laboratory areas occurred in educational properties, 28% occurred in laboratory or science properties and 11% were in manufacturing facilities. Structure Fires in Laboratories, 2/16 1 NFPA, Fire Analysis & Research Division, Quincy, MA
Table 1. Structure Fires in Laboratory Properties, by Area of Origin Area of Origin Fires Civilian Deaths Civilian Injuries Direct Property Damage (in Millions) Laboratory 64 (45%) 0 (NA) 2 (26%) $3.1 (22%) Non-confined 38 (27%) 0 (NA) 2 (26%) $3.1 (22%) Confined 25 (18%) 0 (NA) 0 (0%) $0.0 (49%) Processing or manufacturing area or workroom 10 (7%) 0 (NA) 4 (46%) $0.0 (0%) Non-confined 5 (3%) 0 (NA) 4 (46%) $0.0 (0%) Confined 5 (4%) 0 (NA) 0 (0%) $0.0 (11%) Kitchen or cooking area 8 (5%) 0 (NA) 0 (0%) $0.0 (0%) Non-confined 1 (1%) 0 (NA) 0 (0%) $0.0 (0%) Confined 6 (5%) 0 (NA) 0 (0%) $0.0 (0%) Unclassified technical processing area 4 (3%) 0 (NA) 0 (0%) $0.2 (2%) Non-confined 4 (3%) 0 (NA) 0 (0%) $0.2 (2%) Confined 0 (0%) 0 (NA) 0 (0%) $0.0 (0%) Unclassified equipment or service area 4 (3%) 0 (NA) 0 (0%) $3.1 (22%) Non-confined 3 (2%) 0 (NA) 0 (0%) $3.1 (22%) Confined 1 (1%) 0 (NA) 0 (0%) $0.0 (0%) Office 4 (3%) 0 (NA) 0 (6%) $0.0 (0%) Non-confined 3 (2%) 0 (NA) 0 (6%) $0.0 (0%) Confined 1 (1%) 0 (NA) 0 (0%) $0.0 (0%) Test cell 3 (2%) 0 (NA) 1 (11%) $4.8 (34%) Non-confined 3 (2%) 0 (NA) 1 (11%) $4.8 (34%) Confined 0 (0%) 0 (NA) 0 (0%) $0.0 (0%) Exterior balcony or unenclosed porch 3 (2%) 0 (NA) 0 (0%) $0.0 (0%) Non-confined 0 (0%) 0 (NA) 0 (0%) $0.0 (0%) Confined 3 (2%) 0 (NA) 0 (0%) $0.0 (0%) Other known 42 (30%) 0 (NA) 1 (10%) $2.6 (19%) Non-confined 32 (23%) 0 (NA) 1 (10%) $2.6 (19%) Confined 10 (7%) 0 (NA) 0 (0%) $0.0 (40%) Total 142 (100%) 0 (NA) 9 (100%) $14.0 (100%) Non-confined 89 (63%) 0 (NA) 9 (100%) $14.0 (100%) Confined 52 (37%) 0 (NA) 0 (0%) $0.0 (100%) NA- Not applicable as total deaths were zero. Structure Fires in Laboratories, 2/16 2 NFPA, Fire Analysis & Research Division, Quincy, MA
Note: These are national estimates of fires reported to U.S. municipal fire departments and so exclude fires reported only to Federal or state agencies or industrial fire brigades. Analyses are done separately for non-confined fires (NFIRS incident types 110-123, excluding 113-118) and for fires with incident type codes indicating confined fires (incident type 113-118). Detailed reporting about fire causes and circumstances (such as area of origin) is not required for confined fires. This means there are proportionally far more unknowns to allocate for confined fires. National estimates are projections. Casualty and loss projections can be heavily influenced by the inclusion or exclusion of one unusually serious fire. Fires are rounded to the nearest ten, civilian deaths and injuries to the nearest one, and direct property damage to the nearest million dollars. Damage has not been adjusted for inflation. Estimates include a proportional share of laboratory fires in which the area of origin was unknown. Totals may not equal sums because of rounding error. Source: NFIRS and NFPA fire department experience survey. Structure Fires in Laboratories, 2/16 3 NFPA, Fire Analysis & Research Division, Quincy, MA
Table 2. Structure Fires in Which the Area of Origin Was a Laboratory, by Occupancy 2009-2013 Annual Averages Occupancy Fires Civilian Deaths Civilian Injuries Property Damage (in Millions) Educational 88 (35%) 0 (0%) 17 (71%) $3.2 (24%) College classroom or adult education center 40 (16%) 0 (0%) 4 (16%) $2.4 (18%) High school or middle/junior high 35 (14%) 0 (0%) 13 (53%) $0.0 (0%) Laboratory, utility, defense, or agriculture 73 (29%) 0 (0%) 2 (10%) $4.0 (31%) Laboratory or science laboratory 70 (28%) 0 (0%) 2 (10%) $4.0 (31%) Manufacturing or processing 28 (11%) 0 (100%) 2 (9%) $2.9 (22%) Health care or detention 21 (9%) 0 (0%) 1 (2%) $0.6 (5%) Hospital 12 (5%) 0 (0%) 0 (0%) $0.4 (3%) Mercantile or office 19 (8%) 0 (0%) 2 (6%) $2.1 (16%) Office 12 (5%) 0 (0%) 1 (5%) $1.6 (12%) Residential 9 (4%) 0 (0%) 0 (0%) $0.2 (1%) Public assembly 4 (2%) 0 (0%) 0 (0%) $0.0 (0%) Storage property 3 (1%) 0 (0%) 1 (2%) $0.0 (0%) Unclassified or unknown property use 3 (1%) 0 (0%) 0 (0%) $0.0 (0%) Total 247 (100%) 0 (100%) 24 (100%) $13.0 (100%) Note: These are national estimates of fires reported to U.S. municipal fire departments and so exclude fires reported only to Federal or state agencies or industrial fire brigades. Analyses are done separately for fires not reported as confined fires, for fires reported as chimney or flue, and for fires reported as confined to fuel burner or boiler. Detailed reporting is not required for fires reported as confined fires, which means there are proportionally far more unknowns to allocate for confined fires. Fires reported as confined to cooking vessel, trash container, incinerator, or commercial compactor are not included in the estimates. National estimates are projections. Casualty and loss projections can be heavily influenced by the inclusion or exclusion of one unusually serious fire. Fires are rounded to the nearest ten, civilian deaths and injuries to the nearest one, and direct property damage to the nearest million dollars. Damage has not been adjusted for inflation. Figures reflect a proportional share of home fires with equipment involved in ignition unknown or recorded as heating or air conditioning equipment of undetermined type. Fires reported as no equipment but lacking a confirming specific heat source (codes 40-99) are also treated as unknown equipment and allocated. Home heating fires with this equipment and area of origin unknown have also been allocated proportionally. Totals may not equal sums because of rounding error. Source: NFIRS and NFPA fire department experience survey. Structure Fires in Laboratories, 2/16 4 NFPA, Fire Analysis & Research Division, Quincy, MA
Appendix A. How National Estimates Statistics Are Calculated The statistics in this analysis are estimates derived from the U.S. Fire Administration s (USFA s) National Fire Incident Reporting System (NFIRS) and the National Fire Protection Association s (NFPA s) annual survey of U.S. fire departments. NFIRS is a voluntary system by which participating fire departments report detailed factors about the fires to which they respond. Roughly two-thirds of U.S. fire departments participate, although not all of these departments provide data every year. Fires reported to federal or state fire departments or industrial fire brigades are not included in these estimates. NFIRS provides the most detailed incident information of any national database not limited to large fires. NFIRS is the only database capable of addressing national patterns for fires of all sizes by specific property use and specific fire cause. NFIRS also captures information on the extent of flame spread, and automatic detection and suppression equipment. For more information about NFIRS visit http://www.nfirs.fema.gov/. Copies of the paper forms may be downloaded from http://www.nfirs.fema.gov/documentation/design/nfirs_paper_forms_2008.pdf. NFIRS has a wide variety of data elements and code choices. The NFIRS database contains coded information. Many code choices describe several conditions. These cannot be broken down further. For example, area of origin code 83 captures fires starting in vehicle engine areas, running gear areas or wheel areas. It is impossible to tell the portion of each from the coded data. Methodology may change slightly from year to year. NFPA is continually examining its methodology to provide the best possible answers to specific questions, methodological and definitional changes can occur. Earlier editions of the same report may have used different methodologies to produce the same analysis, meaning that the estimates are not directly comparable from year to year. NFPA s fire department experience survey provides estimates of the big picture. Each year, NFPA conducts an annual survey of fire departments which enables us to capture a summary of fire department experience on a larger scale. Surveys are sent to all municipal departments protecting populations of 50,000 or more and a random sample, stratified by community size, of the smaller departments. Typically, a total of roughly 3,000 surveys are returned, representing about one of every ten U.S. municipal fire departments and about one third of the U.S. population. The survey is stratified by size of population protected to reduce the uncertainty of the final estimate. Small rural communities have fewer people protected per department and are less likely to respond to the survey. A larger number must be surveyed to obtain an adequate sample of those departments. (NFPA also makes follow-up calls to a sample of the smaller fire departments that do not respond, to confirm that those that did respond are truly representative of fire departments their size.) On the other hand, large city departments are so few in number and protect such a large proportion of the total U.S. population that it makes sense to survey all of them. Most respond, resulting in excellent precision for their part of the final estimate. The survey includes the following information: (1) the total number of fire incidents, civilian deaths, and civilian injuries, and the total estimated property damage (in dollars), for each of the major property use classes defined in NFIRS; (2) the number of on-duty firefighter injuries, by type of duty and nature of illness; 3) the number and nature of non-fire incidents; and (4) Structure Fires in Laboratories, 2/16 5 NFPA, Fire Analysis & Research Division, Quincy, MA
information on the type of community protected (e.g., county versus township versus city) and the size of the population protected, which is used in the statistical formula for projecting national totals from sample results. The results of the survey are published in the annual report Fire Loss in the United States. To download a free copy of the report, visit http://www.nfpa.org/assets/files/pdf/os.fireloss.pdf. Projecting NFIRS to National Estimates As noted, NFIRS is a voluntary system. Different states and jurisdictions have different reporting requirements and practices. Participation rates in NFIRS are not necessarily uniform across regions and community sizes, both factors correlated with frequency and severity of fires. This means NFIRS may be susceptible to systematic biases. No one at present can quantify the size of these deviations from the ideal, representative sample, so no one can say with confidence that they are or are not serious problems. But there is enough reason for concern so that a second database -- the NFPA survey -- is needed to project NFIRS to national estimates and to project different parts of NFIRS separately. This multiple calibration approach makes use of the annual NFPA survey where its statistical design advantages are strongest. Scaling ratios are obtained by comparing NFPA s projected totals of residential structure fires, non-residential structure fires, vehicle fires, and outside and other fires, and associated civilian deaths, civilian injuries, and direct property damage with comparable totals in NFIRS. Estimates of specific fire problems and circumstances are obtained by multiplying the NFIRS data by the scaling ratios. Reports for incidents in which mutual aid was given are excluded from NFPA s analyses. Analysts at the NFPA, the USFA and the Consumer Product Safety Commission developed the specific basic analytical rules used for this procedure. The National Estimates Approach to U.S. Fire Statistics, by John R. Hall, Jr. and Beatrice Harwood, provides a more detailed explanation of national estimates. A copy of the article is available online at http://www.nfpa.org/osds or through NFPA's One-Stop Data Shop. Version 5.0 of NFIRS, first introduced in 1999, used a different coding structure for many data elements, added some property use codes, and dropped others. The essentials of the approach described by Hall and Harwood are still used, but some modifications have been necessary to accommodate the changes in NFIRS 5.0. Figure A.1 shows the percentage of fires originally collected in the NFIRS 5.0 system. Each year s release version of NFIRS data also includes data collected in older versions of NFIRS that were converted to NFIRS 5.0 codes. From 1999 data on, analyses are based on scaling ratios using only data originally collected in NFIRS 5.0: NFPA survey projections NFIRS totals (Version 5.0) For 1999 to 2001, the same rules may be applied, but estimates for these years in this form will be less reliable due to the smaller amount of data originally collected in NFIRS 5.0; they should be viewed with extreme caution. Structure Fires in Laboratories, 2/16 6 NFPA, Fire Analysis & Research Division, Quincy, MA
Figure A.1. Fires Originally Collected in NFIRS 5.0 by Year 100% 80% 65% 79% 88% 94% 94% 97% 99% 100% 100% 100% 60% 48% 40% 20% 7% 21% 0% 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 NFIRS 5.0 introduced six categories of confined structure fires, including: cooking fires confined to the cooking vessel, confined chimney or flue fires, confined incinerator fire, confined fuel burner or boiler fire or delayed ignition, confined commercial compactor fire, and trash or rubbish fires in a structure with no flame damage to the structure or its contents. Because this analysis focused on fatalities only, no distinction was made between confined and nonconfined fires. For most fields other than Property Use and Incident Type, NFPA allocates unknown data proportionally among known data. This approach assumes that if the missing data were known, it would be distributed in the same manner as the known data. NFPA makes additional adjustments to several fields. Casualty and loss projections can be heavily influenced by the inclusion or exclusion of unusually serious fire. In the formulas that follow, the term all fires refers to all fires in NFIRS on the dimension studied. The percentages of fires with known or unknown data are provided for non-confined fires and associated losses, and for confined fires only. Rounding and percentages. The data shown are estimates and generally rounded. An entry of zero may be a true zero or it may mean that the value rounds to zero. Percentages are calculated from unrounded values. It is quite possible to have a percentage entry of up to 100% even if the rounded number entry is zero. The same rounded value may account for a slightly different percentage share. Because percentages are expressed in integers and not carried out to several decimal places, percentages that appear identical may be associated with slightly different values. In the formulas that follow, the term all fires refers to all fires in NFIRS on the dimension studied. The percentages of fires with known or unknown data are provided for non-confined fires and associated losses, and for confined fires only. Structure Fires in Laboratories, 2/16 7 NFPA, Fire Analysis & Research Division, Quincy, MA
Area of Origin. Two areas of origin: bedroom for more than five people (code 21) and bedroom for less than five people (code 22) are combined and shown as simply bedroom. Chimney is no longer a valid area of origin code for non-confined fires. Rounding and percentages. The data shown are estimates and generally rounded. An entry of zero may be a true zero or it may mean that the value rounds to zero. Percentages are calculated from unrounded values. It is quite possible to have a percentage entry of up to 100% even if the rounded number entry is zero. The same rounded value may account for a slightly different percentage share. Because percentages are expressed in integers and not carried out to several decimal places, percentages that appear identical may be associated with slightly different values. Structure Fires in Laboratories, 2/16 8 NFPA, Fire Analysis & Research Division, Quincy, MA